Tensor algorithms of blind separation of electromagnetic signals
Dmitry Savostyanov · Russian Journal of Numerical Analysis and Mathematical Modelling · 2010
The problem of blind separation of electromagnetic signals is reduced to joint diagonalization of a set of matrices-statistics. For the solution of this problem we propose complex versions of tensor decomposition algorithms based on the multiplicative update and on the supergeneralized Schur decomposition. A comparison of the obtained results with algorithms of FastICA type is performed and the advantage of tensor methods in the packet signal processing mode is discussed.